A PDF file was originally designed to look the same on every printer and every screen. That design goal, however, makes a lot of everyday PDFs hard for screen reader users to navigate. When a person who is blind or has low vision opens a PDF in a screen reader, the software has to guess what is on the page and in what order it should be read. If the PDF was generated without any structural hints, the reader receives a jumble of words, boxes, and page fragments instead of a clean, logical flow.
If you need this as one file, our free pdf to word tool runs entirely in your browser — nothing is uploaded.
There are three common problems that make ordinary PDFs painful for screen reader users. First, most PDFs have no tag structure. Tags are the metadata that tell assistive technology which part of the document is a heading, which part is a paragraph, which part is a table, and so on. Without tags, the screen reader cannot announce "Chapter Three" before the text of the chapter; it just reads the words in whatever visual order it can reconstruct. Second, even when text is present, the reading order can be scrambled. A two-column layout, a boxed callout, or a figure placed beside its caption can confuse the reader into announcing the caption before the image description, or reading the right column before the left one. Third, a very large share of government forms, school flyers, and nonprofit reports are scanned documents, which means the entire file is really one big image. A screen reader cannot read pixels. If the PDF is a scan with no text layer, a screen reader user gets silence or a single "image" announcement per page, and the document is effectively useless to them.
These three issues are why accessibility checkers report the majority of legacy PDFs as failing. Fixing them is not optional in many public sector contexts; it is a compliance requirement.
Not all PDFs are created equal. The level of accessibility depends largely on how the file was produced. Below is a practical comparison of the three main types you will encounter in real work.
| Type of PDF | How It Is Made | What a Screen Reader Hears | Typical Accessibility Level |
|---|---|---|---|
| Tagged, natively exported PDF | Generated directly from Word or Google Docs with tagging enabled | Headings, paragraphs, tables, and reading order announced clearly | High, if exported correctly |
| Untagged text PDF | Text-based but no semantic tags, often older versions or PDFs generated by print-to-PDF | Words in guessed order, no heading announcements, tables read as one long string | Low to medium |
| Scanned image PDF | A paper document run through a scanner | Only "image" or nothing at all, unless OCR and tagging are added later | None until OCR and tagging are applied |
The key insight is that the first category is the gold standard. When a modern version of Word or Google Docs exports a PDF with tags, the file carries the structural information that screen readers need. The second category is a common middle ground; it has text, so screen readers can read it, but without tags the experience is choppy and the document can be misread. The third category is the worst case. A scanned PDF is essentially a photograph of a page; until you run OCR to create a text layer and then add tags, assistive technology cannot follow the content at all.
Understanding which category your file falls into is the first step. You can usually tell by trying to select text. If you can highlight and copy text, you have at least a text PDF. If you cannot, you are holding a scan.
Whether you are producing a new tagged PDF or remediating an old one, the same set of requirements keeps coming up. Use this list as a quick audit before you publish.
If you can tick all of the boxes above, the PDF will serve most screen reader and low-vision users well. If any box is missing, identify which one and fix it before publishing. In government, school, and nonprofit settings, a single missing requirement can be the difference between a compliant document and a failed audit.
The good news is that the two most common authoring tools can both produce tagged PDFs if you use the right settings. The catch is that the defaults do not always preserve what you need, so you have to check the export dialog.
In Microsoft Word, open your file and go to File, then Save As or Export, and choose PDF as the format. In the save dialog there is an Options button. Inside that dialog, tick the box that says Create tags for accessibility and PDF/A conformance. Without that single tick, Word hands the file over as a clean-looking but untagged PDF. Also make sure your headings are real Heading styles, not bold text that merely looks like a heading; only true heading styles survive the export with semantic meaning. When you are done, open the file in a PDF viewer and confirm that the document structure panel shows headings and regions; that panel is the quickest way to verify that tags came through.
In Google Docs, go to File, then Download, then PDF document. Before the file lands on your desktop, open the print settings and look for a checkbox or option to export with tags and accessibility. Google Docs has improved this area over the last few years, but the option is not always visible by default, so expand the settings section. As with Word, use real heading styles and proper table formatting in the document before you export; the export only carries the structure that already exists in the source file. After you download the PDF, run a quick tag check to make sure the headings and tables survived the round trip.
If your organization has a standard workflow, write the export steps down once and share them with the whole team. Most accessibility failures on new documents trace back to someone skipping the single checkbox that preserves tags.
If you inherit a pile of scanned PDFs, the instinct is to run OCR and call it done. That is the most common mistake in document remediation. OCR adds a hidden text layer under the image so that text can be searched and copied. It does not, by itself, add headings, tag tables, fix reading order, or write alternative text for figures. A screen reader user still gets a rough, unordered dump of words.
The correct order of operations is this. First, run OCR on the scan so that a text layer exists. Use a modern OCR engine and verify that the text layer is accurate by sampling a few pages and comparing it against the image. Second, remap the layout: mark which blocks are headings, which are body paragraphs, and which are tables or figures. This is the tagging step, and it is where the real accessibility is created. Third, add alternative text to every image and mark decorative images as decorative. Fourth, verify the reading order so that the content announces the way a sighted reader would see it. OCR is the entry ticket, but tagging and alternative text are what make the document actually usable by assistive technology. Treat OCR as the first step in a four-step pipeline, not as the finish line.
If your volume of scans is large, budget the tagging time honestly. A single page can take anywhere from a few minutes to over an hour depending on complexity. For high-volume work, some organizations use semi-automated tagging tools and then apply a human review pass; that is the only reliable way to keep quality consistent.
You do not need an external tool to do a useful sanity check. Two quick passes will catch most obvious problems before they reach a user.
The first pass is to open the PDF in a reader that has a built-in accessibility or document structure view. Adobe Acrobat has an Accessibility Check function under the Protect or Tools menu that flags missing language tags, missing alternative text, and broken table structure. Reader and other free viewers also expose a document outline pane; if the outline is empty on a document full of headings, the tags did not survive export. The second pass is to actually read the first two pages out loud with a screen reader, or to open the file in a screen reader if you have one installed on a test machine. Read it the way a user would, by heading navigation first and then linear order. You will hear problems that no visual inspection surfaces, such as captions read before their figures or a table that announces only raw numbers. Two pages is enough to catch the most common structural issues and to confirm that the document's beginning is navigable. If the first two pages pass, the rest of the document is much more likely to pass as well, provided you followed a consistent template.
For a formal check before compliance submission, run a dedicated PDF accessibility checker on the whole file and address every reported error. The goal is not a perfect score; it is a document where no user is blocked from reaching the information.
In the United States, federal agencies and contractors who do business with the government are expected to meet the requirements of Section 508 of the Rehabilitation Act and the related Web Content Accessibility Guidelines that apply to electronic documents, including PDFs. In the European Union, the European Accessibility Act and the Web Accessibility Directive set similar expectations for public sector websites and the documents they publish. You do not need to memorize clause numbers to act on this; you need to know that the standards in force today require tagged structure, alternative text, and a usable reading order for the PDFs you publish, and that a simple untagged export or an unprocessed scan will not meet that bar.
For schools, government offices, and nonprofits, the practical implication is the same: accessibility is now a publishing requirement, not a nice-to-have. Budget the time for tagging and OCR pipelines, train the staff who create documents, and make the checklist above part of the standard review process before anything goes out the door.
Making a PDF work for screen readers is not about adding one extra feature; it is about carrying the structure that was already in your Word document or Google Doc all the way through the export, or rebuilding that structure in a scan that never had it to begin with. The tools are available: the export option in Word and Google Docs, the OCR engine plus a tagging pass for scans, and the accessibility check in Acrobat and other readers. What separates a compliant document from a broken one is the discipline of running through the checklist every single time, and the willingness to read your own document through a screen reader at least once before you publish it. If you can do that, the rest of the document will follow.
For teams that move files between formats regularly, the workflow helpers at /tools/pdf-to-word/ and /tools/word-to-pdf/ are worth keeping in your toolkit. Converting a legacy PDF back into an editable format lets you restructure it properly before re-exporting a tagged version; the reverse direction is how most tagged PDFs are born in the first place. Both directions matter, and both should be part of a repeatable accessibility pipeline rather than one-off fixes.
How do I add alternative text to images in my PDF?
Select the image and choose "Add Alt Text" from the tools menu. Describe the visual content concisely, focusing on meaning rather than appearance. For decorative graphics, leave the text field empty to signal screen readers to skip it. Save the document to apply changes, ensuring the metadata is embedded correctly.
What is the best way to organize headings in my document?
Use the "Tag Document" feature to assign Heading Levels 1 through 6 to titles and subtitles. Ensure each section has a clear hierarchy. Avoid using font size or bold text as the only indicator, as screen readers rely on semantic tags. This creates a navigable outline for users exploring the structure.
Does reading order matter for complex layouts like columns?
Yes, reading order must match the logical flow of content. In multi-column layouts, screen readers may mix text if tags are missing. Use the "Set Reading Order" tool to define the sequence explicitly. This prevents jumbled text when navigating and ensures users hear information in the correct context without confusion.
Can I fix broken links in an existing PDF?
Enable the "Link" tool to identify unlinked text that should be interactive. Select the text and paste the URL to create a valid anchor. Test each link to confirm it opens correctly. Properly formatted links allow screen readers to announce the destination, improving accessibility for users relying on keyboard navigation.